On Mon 09-10-23 17:45:57, Lizhi Xu wrote:
> When sbi->flag is JFS_NOINTEGRITY in lmLogOpen(), log->bdev_handle can't
> be inited, so it value will be NULL.
> Therefore, add the "log ->no_integrity=1" judgment in lbmStartIO() to avoid such
> problems.
>
> Reported-and-tested-by: syzbot+23bc20037854bb335d59@xxxxxxxxxxxxxxxxxxxxxxxxx
> Signed-off-by: Lizhi Xu <lizhi.xu@xxxxxxxxxxxxx>
Ah, good catch. Who would think someone creates bios for NULL bdev only to
release them shortly afterwards ;). Anyway the fix looks good. Feel free to
add:
Reviewed-by: Jan Kara <jack@xxxxxxx>
Christian, please pick up this fixup into your tree. Thanks!
Honza
> ---
> fs/jfs/jfs_logmgr.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/fs/jfs/jfs_logmgr.c b/fs/jfs/jfs_logmgr.c
> index c911d838b8ec..c41a76164f84 100644
> --- a/fs/jfs/jfs_logmgr.c
> +++ b/fs/jfs/jfs_logmgr.c
> @@ -2110,10 +2110,14 @@ static void lbmStartIO(struct lbuf * bp)
> {
> struct bio *bio;
> struct jfs_log *log = bp->l_log;
> + struct block_device *bdev = NULL;
>
> jfs_info("lbmStartIO");
>
> - bio = bio_alloc(log->bdev_handle->bdev, 1, REQ_OP_WRITE | REQ_SYNC,
> + if (!log->no_integrity)
> + bdev = log->bdev_handle->bdev;
> +
> + bio = bio_alloc(bdev, 1, REQ_OP_WRITE | REQ_SYNC,
> GFP_NOFS);
> bio->bi_iter.bi_sector = bp->l_blkno << (log->l2bsize - 9);
> __bio_add_page(bio, bp->l_page, LOGPSIZE, bp->l_offset);
> --
> 2.25.1
>
--
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR
